---
title: NotebookLM Project Workflow — Document Upload & Interaction
type: article
created: '2025-09-30'
updated: '2025-09-30'
source_docs:
- raw/2025-09-30-ops-sync-90775978.md
tags:
- ai-tools
- notebook-lm
- workflow
- content-creation
- knowledge-management
layer: 2
client_source: null
industry_context: null
transferable: true
---

# NotebookLM Project Workflow — Document Upload & Interaction

Google's NotebookLM is a document-grounded AI tool that constrains its responses to the sources you provide — making it ideal for client work where you need accurate, on-brand output without AI hallucination or off-topic content.

See also: [[wiki/knowledge/ai-tools/ai-tool-selection-guide]] | [[wiki/knowledge/ai-tools/claude-projects-workflow]] | [[wiki/knowledge/ai-tools/prompting-best-practices]]

---

## What NotebookLM Is Good For

- Synthesizing large volumes of client-provided documents (emails, briefs, decks, research)
- Generating summaries, mind maps, and in-depth reports grounded in your source material
- Ensuring AI output stays within what you actually know about a client — no fabrication
- Early-stage project onboarding when you need to get up to speed quickly on a topic

**Key differentiator:** Unlike ChatGPT or Claude in a standard chat session, NotebookLM will not go out to the internet or invent information. It only works from what you give it.

---

## Step-by-Step Workflow

### 1. Create a New Notebook

1. Go to [notebooklm.google.com](https://notebooklm.google.com)
2. Click **New Notebook**
3. Give it a descriptive name (e.g., `Blue Sky Capital — Content Project`)

### 2. Upload Your Source Documents

Feed it everything you have about the project or client. Supported sources include:

| Source Type | How to Add |
|---|---|
| PDFs | Upload directly |
| Google Docs | Connect via Google Drive |
| Plain text / Markdown | Upload or paste |
| Website URLs | Paste the link directly |
| Emails | Copy into a Google Doc, then upload |
| PowerPoint / Slides | Export to PDF first, then upload |

> **Tip:** If a client sends emails, paste them into a single Google Doc and upload that. Don't try to upload raw `.eml` files.

NotebookLM supports up to ~200 source documents per notebook.

### 3. Let NotebookLM Orient Itself

After uploading, NotebookLM will generate a brief summary of what it understands from your sources. Read this carefully — it tells you whether the tool has correctly understood the project scope. If something is missing or wrong, add more documents before proceeding.

### 4. Interact with the Notebook

Once your sources are loaded, use the chat interface to ask questions and generate content. The tool will only draw on your uploaded materials.

**Example prompts to get started:**
- *"Summarize the key services this company offers based on the documents I've provided."*
- *"What are the main themes across these client emails?"*
- *"Draft a content brief for a blog post about [topic] using only what's in these documents."*

### 5. Use Built-In Features

- **Mind Map** — Generates a visual overview of topics across your sources. Useful for spotting gaps or structuring content.
- **In-Depth Report** — Produces a longer synthesis document from your sources.
- **Audio Overview** — Generates a podcast-style summary (useful for quick review).

---

## When to Use NotebookLM vs. Other Tools

| Situation | Best Tool |
|---|---|
| You have client documents and want grounded output | **NotebookLM** or [[wiki/knowledge/ai-tools/claude-projects-workflow\|Claude Projects]] |
| You need internet research with citations | [[wiki/knowledge/ai-tools/perplexity-guide\|Perplexity]] |
| You need creative writing or reasoning | Claude or ChatGPT |
| You need Google Sheets/Docs output | Gemini |
| You want to cross-check AI output | Run output through a second tool (e.g., Claude reviewing ChatGPT's draft) |

---

## Practical Tips

- **More context = better output.** Don't just upload one document. The more relevant material you provide, the more accurate and useful the responses will be.
- **Verify the notebook "knows" the topic.** After uploading, ask it a basic question about the client or subject before asking it to write anything. Confirm it understood correctly.
- **Convert before uploading.** PowerPoint files should be exported to PDF. Emails should be pasted into a Google Doc.
- **NotebookLM is not a replacement for reading the material.** You still need to review AI-generated output and be able to speak to it — especially before client presentations.

---

## Limitations

- Cannot browse the internet or pull in real-time data
- Does not generate Google Sheets or Docs natively (use Gemini for that)
- Mind maps can be sparse if source documents are thin
- Maximum ~200 source documents per notebook

---

## Origin

This workflow was documented during an internal ops sync on 2025-09-30, where Mark Hope walked through NotebookLM and related AI tools as part of a broader discussion on improving content quality. A formal team professional development session covering these tools is planned for October. See [[wiki/meetings/2025-09-30-ops-sync]] for full context.